Accuracy of measurement while drilling/logging while drilling (MWD/LWD)
depth measurements can be improved by considering the dynamic variation in
drillstring length caused by pipe loading under changing drilling conditions.
This paper details a new method that uses surface torque, hookload, and
temperature measurements to determine force distribution in a drillstring and
to compute apparent drillstring length. When available, torque, weight on bit
(WOB), and temperature measured downhole are used to increase accuracy and
robustness of the method.
Although logging depth is referred to as a measurement, in reality only the
drilling block position is measured. Depth is inferred from it using
drillstring length. In recent publications, physical phenomena affecting this
were analyzed and quantified. Elastic pipe stretch and thermal expansion were
found to be most significant. Techniques to compensate for these effects on the
basis of empirical formulae have been proposed (Brooks et al. 2005), but they
provide an averaged correction that has insufficient accuracy for many drilling
and formation evaluation applications.
This paper presents experiments covering various wellbore profiles,
temperature profiles, and drilling modes, which show that the depth fluctuation
may be as much as 2.7 meters with a 7,000–meter (m) long drill string even when
only the current rig operation mode changes. Among other factors considered in
the paper, apparent depth fluctuation is the most significant contributor to
commonly observed MWD/LWD log discrepancies when bed boundaries or other
features are not logged at the same depth with each sensor. These errors lead
to inaccurate petrophysical calculations, distortion of borehole images, and
lost time caused by depth matching. Case studies illustrate the positive effect
of dynamic depth correction on formation evaluation log quality.
The accuracy of a depth measurement is normally estimated in terms of its
bias and uncertainty. A significant portion of the depth bias is caused by
elastic stretch and thermal expansion (for example in a 7,000-m long vertical
drillstring they can be 9 m and 6 m, respectively). The proposed method removes
this bias and allows improved depth uncertainty.
Measured depth uncertainty (1 sigma) in the Industry Steering Committee on
Wellbore Surveying Accuracy (ISCWSA) MWD model caused by drill string stretch
is 2.2 X 10–7 m–1, multiplied by measured depth (MD) and by true vertical depth
(TVD). Uncertainty in measured depth cannot be completely eliminated even by
applying corresponding corrections because of the modeling and input data
inaccuracies. Nevertherless, it is estimated that the proposed method
significantly reduces this uncertainty (e.g., 50% and more, depending on the
wellbore, available data, etc.).
Improved depth accuracy, in turn, reduces the uncertainty in computation of
reservoir characterization parameters, such as net-to-gross and structural dip,
especially when data from multiple wells are evaluated together.
Depth is one of the most important formation evaluation measurements, but
one of the most difficult to define accurately. Previous publications (Wilson
et al. 2004; Brooks et al. 2005; Pederson and Constable 2006) detail this
problem as occurring to various degrees for both wireline- and drillpipe-based
systems. With longer and deeper wells in deeper provinces around the world, and
the use of drill pipe conveyance (MWD/LWD), this problem becomes more
Awareness of the financial as well as technical implications for inaccurate
depth is increasing. Depth accuracy is also vital for accurate calculation of
structural dip from borehole images, picking perforation points, and
correlation of geological units. MD is used directly in the calculation of
TVD—the primary depth used for reservoir delineation. Wilson et al. (2004)
explore one such case in which a 2-m TVD discrepancy in oil water contact (OWC)
has a widespread implication for the field development plan, pressure support,
and compartmentalization with a significant cost attached.
Errors in depth are difficult to detect using a single scalar measurement,
however, comparison of multiple curves, which have similar character (Fig.
1) and array measurements, often readily exhibit these artifacts. It is
common practice to use a cross-correlation and depth “rubber-banding” technique
to bring MWD/LWD measurements from different sources (even within the same
bottomhole assembly [BHA]) together, using measurements from the sensor closest
to the bit as a reference. This practice addresses the symptom but not the
cause, with no reliable reference or rationale (e.g., the sensor closer to bit
generally is not more accurate). Misalignment of curves serves as a good
indicator of the formation evaluation (FE) measurements depth placement
uncertainty. It should rather be used for validation of methods used to convert
LWD/MWD data to depth logs.
This paper discusses the source of the marked relative variations between FE
measurements and concentrates on improvement in the accuracy of
drillstring-derived depth for MWD/LWD, which leads to smaller uncertainties
around the calculation of final TVD.
The integrity of the depth measurement can be described by a variety of
terminologies, generally depending on the application of depth under
discussion. Wellbore placement, tying in formation evaluation data with seismic
sections, and crosswell correlation all rely on accurate absolute depth (both
measured and calculated TVD). Depth accuracy is the degree of conformity of the
measured value to the true value.
© 2008. Society of Petroleum Engineers
View full textPDF
- Original manuscript received:
4 October 2006
- Meeting paper published:
24 September 2006
- Revised manuscript received:
4 June 2007
- Manuscript approved:
8 September 2007
- Version of record:
20 March 2008